Constructor. :param epsilon: epsilon to project on :type epsilon: float

Constructor. :param log_dir: summary directory :type log_dir: str

Constructor. :param writers: writers :type: [SummaryWriter]

Constructor. :param log_dir: summary directory :type log_dir: str

Constructor. :param log_dir: summary directory :type log_dir: str

Constructor. :param log_dir: summary directory :type log_dir: str

Constructor. :param data: summary directory :type data: str

Constructor. :param log_dir: summary directory :type log_dir: str

Constructor.

Constructor. :param message: message :type message: str :param level: level :type level:

Constructor. :param model: model :type model: dict or torch.nn.Module :param optimizer: optimizer :type opti

Constructor. Exactly one of include_list and exclude_list must be specified. :param img_dims: image dimensions :type img_dim

Constructor.

Constructor. :param shape: shape :type shape: [int]

Constructor.

Constructor. :param n_channels: number of channels :type n_channels: int

Constructor. :param images: images/inputs :type images: str or numpy.ndarray :param labels: labels :type lab

Constructor. :param images: images :type images: str or numpy.ndarray :param perturbations: additive perturbations

Initialize and start timer.

Constructor.

Constructor.

Constructor.

Initialize. :param config: configuration :type config: [str]

Initialize. :param config: configuration :type config: [str]

Initialize. :param config: configuration :type config: [str]

Initialize. :param target_config: configuration :type target_config: [str] :param attack_config: configuration

Constructor. For more details for confidence-thresholded evaluation, see adversarial_evaluation.py. :param clean_probabilit

Setup adversarial evaluation. Here, the following dimensions are expected/supported: * `clean_probabilities`: predicted pro

Constructor. For more details for confidence-thresholded evaluation, see adversarial_evaluation.py. :param probabilities: p

Constructor. For more details for confidence-thresholded evaluation, see adversarial_evaluation.py. :param clean_probabilit

Constructor. :param model: model :type model: torch.nn.Module :param trainset: training set :type trainset:

Constructor. :param writer: summary writer :type writer: torch.utils.tensorboard.SummaryWriter or TensorboardX equivalent

Constructor. :param model: model :type model: torch.nn.Module :param trainset: training set :type trainset:

Constructor. :param model: model :type model: torch.nn.Module :param trainset: training set :type trainset:

Initialize. :param args: optional arguments if not to use sys.argv :type args: [str]

Initialize. :param args: optional arguments if not to use sys.argv :type args: [str]

Initialize. :param args: optional arguments if not to use sys.argv :type args: [str]

Initialize classifier. :param N_class: number of classes to classify :type N_class: int :param resolution: resolutio

Initialize classifier. :param N_class: number of classes to classify :type N_class: int :param resolution: resolutio

Constructor. :param inplanes: input channels :type inplanes: int :param planes: output channels :type planes

Constructor. :param in_planes: number of input planes :type in_planes: int :param planes: output planes :typ

Initialize classifier. The keyword arguments, resolution, number of classes and other architecture parameters from subclasse

Constructor. :param in_planes: number of input planes :type in_planes: int :param planes: output planes :typ

Initialize classifier. :param N_class: number of classes to classify :type N_class: int :param resolution: resolutio

Constructor. :param inplanes: input channels :type inplanes: int :param planes: output channels :type planes

Initialize classifier. :param N_class: number of classes to classify :type N_class: int :param resolution: resolutio

Initialize classifier. :param N_class: number of classes :type N_class: int :param resolution: resolution :t

Initialize classifier. :param N_class: number of classes to classify :type N_class: int :param resolution: resolutio

L_0 corner search attack. :param max_iterations: iterations :type max_iterations: int :param epsilon: epsilon :type epsilon: flo

L_0 normalized PGD implementation. :param base_lr: learning rate :type base_lr: float :param max_iterations: iterations :type ma

L_0 normalized PGD implementation with zero initialization. :param base_lr: learning rate :type base_lr: float :param max_iterations

L_0 corner search attack. :param max_iterations: iterations :type max_iterations: int :param epsilon: epsilon :type epsilon: flo

L_1 normalized PGD implementation. :param base_lr: learning rate :type base_lr: float :param max_iterations: iterations :type ma

L_1 normalized PGD implementation with zero initialization. :param base_lr: learning rate :type base_lr: float :param max_iterations

L_2 cube attack. :param max_iterations: iterations :type max_iterations: int :param probability: probability of change :type pro

L_2 normalized PGD implementation. :param base_lr: learning rate :type base_lr: float :param max_iterations: iterations :type ma

L_2 normalized PGD implementation with zero initialization. :param base_lr: learning rate :type base_lr: float :param max_iterations

L_inf normalized PGD implementation with zero initialization. :param base_lr: learning rate :type base_lr: float :param max_iteratio

L_inf query limited attack. :param base_lr: learning rate :type base_lr: float :param max_iterations: iterations :type max_itera

Print network.

L_inf query limited attack with zero initialization. :param base_lr: learning rate :type base_lr: float :param max_iterations: itera

Add given attacks to given list of attack variables. Allows to keep an overview. :param config_attack_variables: list of attack config names

Add figure. :param tag: tag :type tag: str :param figure: test :type figure: matplotlib.pyplot.figure

Add figure. :param tag: tag :type tag: str :param figure: test :type figure: matplotlib.pyplot.figure

Add figure. :param tag: tag :type tag: str :param figure: test :type figure: matplotlib.pyplot.figure

Add histogram data. :param tag: tag :type tag: str :param values: values :type values: torch.Tensor or numpy

Add histogram data. :param tag: tag :type tag: str :param values: values :type values: torch.Tensor or numpy

Add histogram data. :param tag: tag :type tag: str :param values: values :type values: torch.Tensor or numpy

Add image. :param tag: tag :type tag: str :param img_tensor: image :type img_tensor: torch.Tensor or numpy.n